Tuesday, May 21, 2013

Why I left Heroku, and notes on my new AWS setup

Why I left Heroku, and notes on my new AWS setup

to spread the word about how we did it and why you should consider it if decided to use it for a few reasons: * Heroku had the promise of easy setup and easy scaling in cases of high traffic. up my app -- and this auto-detection stuff is the kind of thing that's very I verified that the server could send email, and all the necessary but I kept with it. after I scaled up the number of dynos. Whenever I tried to deploy, I got [ugly error Then, one April evening, I deployed my app, and Heroku decided to upgrade the down, after I had just sent the link to some potential partners who, for all I then on, whenever I deployed, I got a little nervous that something bad would Heroku offers a [way to Heroku for this blog, and I might use it in the future for small/throwaway the years, I had no idea how powerful Amazon's full suite of services really I also converted the After that instance had all my code/dependencies on it, I created an AMI from it ("Create Image (EBS AMI)" in the EC2 dashboard). Basically, Amazon constantly monitors the app servers, and if any of them reaches a certain CPU usage, Amazon will automatically launch X new server(s) and associate them with the load balancer when they're up and running Rather than have to deal with that, I changed the app to use cookie-based sessions, so that session data is stored in signed cookies rather than in memcache I fell in love with it the moment I scaled it from one to two availability zones with a couple of clicks on the AWS admin console See here for the relevant parts of my fabfile. Then I associated the load balancer with the new AMI, and each new app server from then on will use the new AMI Puppet to automatically install the necessary is just as easy to use as Heroku (once it's set up!), with the full power of

Source: http://www.holovaty.com/writing/aws-notes/

No comments:

Post a Comment

Share this!